Analytical Reproducibility of the Breast Cancer Intrinsic Subtyping Test and nCounter Analysis System Using Formalin-Fixed Paraffin-Embedded (FFPE) Breast Tumor Specimens
Torsten Nielsen, Sandra McDonald, Shashikant Kulkarni, James Storhoff, Carl Schaper, Brett Wallden, Sean Ferree, Shuzhen Liu, Vishwa Hucthagowder, Katherine Deschryver, Vicky Holtschlag, Garrett Barry, Michael Evenson, Naeem Dowidar, Malini Maysuria, Doris Gao. British Columbia Cancer Agency, Vancouver, BC, Canada; Washington University School of Medicine, St. Louis, MO; NanoString Technologies, Seattle, WA; MyRAQA Inc., Redwood Shores, CA
Background: NanoString's breast cancer intrinsic subtyping test is based on the previously reported PAM50 gene expression signature. The test is performed on the nCounter Analysis System using gene-specific fluorescently labeled probe pairs that hybridize directly to target mRNAs in solution. The intrinsic subtype (Luminal A/B, Her2-enriched, Basal-like), ROR score (0-100 scale), and risk category are determined by the system using a prospectively defined and locked algorithm. A recent clinical validation in over 1000 patient specimens from the ATAC trial demonstrated that the ROR score added significant prognostic information beyond the Oncotype DX® RS score in estimating the likelihood of distant recurrence in hormone receptor positive, post-menopausal breast cancer patients. The validation studies described here were designed to measure the analytical robustness of the test across three clinical testing sites.
Design: Analytical precision was measured by testing five pooled breast tumor RNA samples representing each breast cancer subtype and risk classification group across three sites, six operators, and three reagent lots. Each site completed 18 valid runs consisting of 10 tests each. Reproducibility was measured by testing replicate tissue sections from 43 FFPE breast tumor blocks following independent review of an H&E stained slide by a pathologist at each site to mark the area of invasive carcinoma. Following macrodissection of tumor tissue, total RNA was isolated and 125 – 500 ng RNA was tested on the nCounter system.
Results: Within each RNA sample, the measured SD was less than 1 ROR unit including all sources of variation, and there were no significant differences in performance by operator or site. In the reproducibility study, the measured total SD of the ROR score across all samples was 2.9 ROR units. The average site to site concordance of risk category was > 90 %, and there were no low-to-high risk mis-classifications (or vice versa). The average site to site subtype call concordance was 97 %.
Conclusions: The nCounter system provides a highly precise method for measuring clinically-relevant gene expression signatures on FFPE surgical pathology specimens, and the analytical reproducibility of the test has been validated using FFPE breast tumor specimens across multiple clinical testing laboratories.
Wednesday, March 6, 2013 9:30 AM
Poster Session V # 262, Wednesday Morning